Application of Cutting-Edge 3D Seismic Attribute ... · PDF fileApplication of Cutting-Edge 3D Seismic Attribute Technology to the Assessment of Geological Reservoirs for CO ... Geology

Liner, U. Houston 3D Seismic Attributes and CO2 Sequestration

1

Application of Cutting-Edge 3D Seismic Attribute Technology to the

Assessment of Geological Reservoirs for CO2 Sequestration Type of Report: Progress Frequency of Report: Quarterly Reporting Period: January 1, 2010 – March 31, 2010 DOE Award Number: DE-FG26-06NT42734 [University Coal Research]

(UH budget G091836) Submitting Organizations: Department of Earth and Atmospheric Sciences

Reservoir Quantification Lab University of Houston Houston, Texas 77204-5505

Preparers: Prof. Christopher Liner - P.I.

Dr. Jianjun (June) Zeng (coordinator for this report) Dr. Po Geng Heather King Phone: 713-743-919 Fax: 713-748-7906


2

CONTENTS

Executive Summary ............................................................................................................ 3

Geology and Geophysics ................................................................................................ 4 Reservoir Simulation ...................................................................................................... 7

Work Plan for the Next Quarter........................................................................................ 10 Cost and Milestone Status................................................................................................. 11 Summary of Significant Events ........................................................................................ 11 Project Personnel .............................................................................................................. 11 Technology Transfer Activities ........................................................................................ 12 Contributors ...................................................................................................................... 12 Appendix A: Relative Permeability Equations ................................................................. 14 Tables................................................................................................................................ 16 Figures............................................................................................................................... 18


3

Executive Summary

This was the last quarter of the Dickman project and efforts focused on finalizing the geologic gridded property model and flow simulation grid to achieve a satisfactory history match. The latter is critical for validation of the gridded property model. Work has already begun on the final project report (due 30 June 2010), with various components of the report allocated to researchers associated with the project. The project final report outline and content will be designed in collaboration with our DOE program officer.


4

Activities in Quarter Activities in this quarter are focused on the final report on the scientific achievements and

effective workflow of the Dickman CO2 sequestration project. The planned contents,

authors and progresses are shown in Table 1. In addition, efforts were made to tackle the

remaining challenges that may become a part of the final report, or further focus of the

Dickman research. Three major tasks were performed in geology and geophysics: 1)

Detailed re-interpretation of the Mississippian horizon spice volume to compare with

amplitude interpretation, 2) Re-adjustment of input reservoir properties based on the

results of history matching flow simulation, 3) Up-dated information on regional

stratigraphy and tectonic history based on new Kansas Geological Survey (KGS)

publications through 2009. The first task resulted in a paper submitted to the upcoming

SEG 2010 and a poster presentation at the Annual DOE Pittsburgh CCS Conference in

May 2010. In flow simulation research, two rounds of history matching simulations were

performed on Dickman filed production data to yield good matches in oil and water

production. Moreover, an optimized input parameter set was obtained by analyzing misfit

causes. This further validated the functionality and the efficiency, as well as risks, in

using the CMG black oil simulator to predict CO2 injection.

Geology and Geophysics

Comparison of spice and amplitude interpretation. Spice is an attribute computed on

migrated stack traces using the continuous wavelet transform and the theory of

singularity analysis (Li and Liner, 2008). It has a higher apparent resolution than

amplitude, and therefore is useful when trying to map subtle features. Figure 1 shows the

top Mississippian time structure map from tracking amplitude (A) and spice (B). At the

top Mississippian, a zero crossing in amplitude was auto-tracked as well as the

corresponding peak in spice. Both events were tracked first in 8 intersecting 2D lines,

with a 2 ms window, followed with 2 iterations of standard 3D tracking. The spice map

has better continuity across the mapped area with fewer tracking gaps (Figure 2). In the

difficult channel zone, spice gives a better time structure picture. Tracking an event in

spice seems much more helpful for distinguishing small scale features than tracking the

corresponding event in an amplitude volume.


5

Since spice peaks correspond to amplitude zero crossings, it follows that when spice

shows apparent thin beds the zero crossings are closer together, and this must relate to a

local increase in data frequency. One such area in our data is near the Miss unconformity

(Figure 1D). Spice and instantaneous frequency are shown for data along the line in

Figure 3. As expected, the frequency is often anomalous (high or low) at the Miss.

unconformity surface and could be used as a secondary attribute for mapping. However,

frequency shows little vertical detail compared to spice and certainly could not be used as

a substitute. Also shown in Figure 3 are negative curvature (C) and positive curvature

(D). Much like coherence, curvature attributes (Marfurt, 2006) are computed with an

extended operator, while spice is a point-wise computation localized in both space and

time. In this case, curvatures were computed with a 20 ms time window (red bar).

Setting aside the vertical resolution mismatch, curvature and spice both indicate an

anomalous area between the Mckinley A-1 and Elmore 3 wells in the vicinity of the Miss

unconformity (Nissen, et al., 2006). Curvature and spice are complimentary attributes

that can be used to improve detailed interpretation.

Flow model adjustment based on history matching. After analysis of results from the

first round of history matching runs, several adjustments were made to improve the fit.

The Dickman field is a dominantly carbonate reservoir, with a large oil-water transition

zone associated with most of the oil production. Flow grid models resolving the

transition zone were developed for this reason.

The input reservoir property grid for the flow simulation was also calibrated based on the

first round results. The propagation of permeability within the 3D grid was redone using

325 degree azimuth as a preferred direction, roughly parallel to the NW-trending

fractures that were considered as open fractures (Nissen, et al., 2006).

Connate water saturation is an important input parameter for the simulation. In Dickman

reservoir core analysis data, connate water saturation (0.2-0.7) seems too high. This water

saturation was measured on flushed cores, therefore is likely much higher than in-situ

water saturation. Statistics relating the measured water saturation from flushed cores and


6

the in-situ water saturation were used (Figure 4) to re-scale the measured connate water

saturation. The revised values (0.2-0.3) represent a good approximation to the Dickman

core data. A value of 0.2 was used in the final simulation runs.

Updated regional stratigraphy. The Kansas Geological Survey has recently updated

the regional stratigraphic chart (Figure 5) for Kansas (Sawin, 2008, 2009). We have

synchronized the local stratigraphy at the Dickman field to the new regional chart (Figure

5), including the project target interval of Ft. Scott to top Viola (‘This Study’ blue box in

Figure 5). The purpose of this synchronization is to reconstruct a regional structural

deformation history that may have controlled the faulting and fracturing events in the

target strata.

In the studied area, a part of the younger strata including the Upper Cretaceous and

Tertiary is exposed on the steep slopes of the river valleys. The surface exposures include

the Upper Cretaceous sandy and chalky shale, and overlying Tertiary unconsolidated or

consolidated sand, shale and gravel.

Over 4,900 ft of the older strata were penetrated in the studied area and have been

correlated, some with uncertainties, to the regional stratigraphic column established

mainly on outcrop studies. The subsurface strata overlaying the Pre-Cambrian basement,

from the oldest to the youngest are as follows: Undifferentiated Ordovician/Cambrian

and Ordovician dominated by carbonates, Mississippian carbonates, Pennsylvanian cyclic

carbonates and clastic rocks, Permian red-bed secessions, and Lower Cretaceous shale

and sandstones with chalky beds. The general lithology of these units, especially those

taken as correlation keys in both well logs and seismic profiles, will be described in detail

for the final report.

The reconstruction of major structure deformation events is based mainly on published

studies (Blakely, 2004; Merriam, 1963;). Well tops in the KGS database on the

southwest side of the Central Kansas Uplift (CKU) are also used to trace deposition,


7

deformation, and preservation of strata as evidence of structural activities. The movement

of the CKU controlled the local faulting and fracturing.

After the major post-Miss. and Pre-Penn. structural movement related to the continental

collision along the Ouchita mountain belt, there were at least two structural events that

may have left significant footprints in the studied area. The first event was during the late

Pennsylvanian time, as indicated by significant thinning of the Upper Penn. Lansing

group on top of the CKU and the abrupt thickness changes (up to 200 ft) of Lansing and

Kansas City groups along a NE-trending lineation. This lineation is parallel to the south

boundary of the CKU. Since Lansing group conformably underlays the younger Penn.

strata (Zeller et. all, 1968), these thickness changes suggest syn-depositional uplift of the

CKU and local subsidence along the NW side. This event might influence the faulting

and fracturing in the Dickman area, such as the NE-trending boundary fault and a couple

of NW-trending faults to the north end of the survey area. The second structural event is

during late Cretaceous or later, probably associated with the Laramide Orogeny. It

results in secondary structures, such as the Northeast trending Aldrich Anticline seen in

the Eldritch Northeast field of the studied area. They are perpendicular to the axis of the

CKU, and are probably the result of uplift and adjustment caused by stresses along pre-

existing zones of weakness (Ramaker, 2006). These anticlines are likely associated with

strike-slip movements. As a result, the northeast trending boundary fault in the studied

area may have been closed to become a sealing fault to the pay zones in Dickman,

Humphrey and Sargent field areas.

Reservoir Simulation This quarter has seen a major achievement in reservoir simulation, the completion of oil

and water history matching for the Dickman field. History-match simulation is a very

challenging task and has been a weak point in the CO2 injection simulation, even in

recent publications. History-matching work for oil production in the Dickman filed

(based on 15 wells), provided important information on methods, input parameter

selection, optimization, and risk assessment for the further geological and flow modeling

in CO2 sequestration researches.


8

The conventional role of history matching is to validate and calibrate the reservoir model.

The reservoir properties, formation structural data, and production data collected from

geological model analysis and the production company (Grand Mesa) can only be

validated and calibrated through the history matching process. In addition, the shallow

geologic section forms the cap rock for the deep saline aquifer system which is our CO2

sequestration target. A good understanding of the shallow section is essential for safe

CO2 storage in deep saline aquifers. Dickman field history matching will also provide us

a good opportunity to understand the shallow section integrity.

A major challenge for the history-matching work is lack of necessary information from a

mature field like Dickman, which was discovered and put into production in 1962. From

Hilpman et al. (1964), Carr (2006) and well log data, we collected the following field

data:

Acreage = 240 acres

Net Pay Zone Thickness = 7 feet

Average depth = 4424 feet in TVD

Oil API gravity = 37 API (0.84 g/cm3)

The reservoir Temperature = 113 oF

The reservoir average pressure = 2066 psia

TDS (Total Dissolved Solid) salinity = 45,000 ppm

The aquifer water density at reservoir condition

The reservoir water compressibility at reservoir condition

Oil Water Contact (OWC) = 4578 ft in TVD.

The pressure-volume-temperature constant at thermodynamic equilibrium (PVT data) is

used to determine the volume ratios of oil and gas. The volume ratios relate the density at

reservoir conditions to the density at surface conditions, therefore the in-situ volumes

under formation temperature and pressure determine produced volumes at surface


9

conditions. Since PVT data were not available in this study, it had to be evaluated by

correlations. The Computer Modelling Group (CMG) software package for black oil, gas

and water PVT (McCain, 1991) was used to predict fluid properties. The correlation to be

used is determined by an API gravity criterion (Lasater if API > 15, otherwise Standing).

API value for Dickman field is 37, so Lasater’s correlation was used.

In a multiple phase flow system containing oil, gas and water, the relative permeability is

a function of phase saturation. Stone correlation formulations were used to generate the

relative permeability model for relative permeability of water, oil and water, gas and

water, and gas (equations in Appendix A).

Figure 6 shows the simulation grid with 15 production wells and one injection well.

Table 2 lists production starting date, ending date, and water break through time for all 15

production wells. As shown in Table 2, there is complete oil production data record for

all production wells. However, we have complete water production records only for five

wells: Dickman 1, Humphrey 2, Humphrey 3, Humphrey 4 and Tilley 5. For the

remaining production wells, there is either no water production data or only partial data.

Dickman 4 is the water injection well which is used to inject the field waste water back

into the reservoir. Because the water injection data record is not available, we assumed

that Dickman 4 injected all produced water back to the reservoir.

The relative permeability model, well perforation, net pay zone thickness, PVT and

capillary pressure model are the reservoir properties being calibrated to match the water

and oil production data. These properties have been updated after each of the simulation

run.

In the final simulation, the original PVT model generated by Lasater’s correlation was

used, and generated good matching results. There was no gas or condensate recorded in

Dickman field production reports. The compressibility of water is very small, and the

effect of gas compressibility on the production volumes of oil and gas is not very

significant.


10

A good oil and water history match was obtained on the production wells around the

Dickman 4 injector, Dickman 1 being a good example (Figure 7). Dickman 1 is the first

production well in the Dickman field and is still under production at present day.

Matching the Dickman 1 production is an important step in modeling the entire field.

Results also suggest that one main function of the Dickman 4 water injection well is to

maintain reservoir pressure. Without it the reservoir pressure would be below 1000 psi at

the end of the history matching simulation. The assumption that Dickman 4 injected all

produced water back into the reservoir seems to be correct.

Well perforation data also plays an important role in history matching. Dickman A1 was

reportedly perforated only in the two upper flow model layers (uppermost Mississippian),

and the simulated oil production rate is consequently much lower than the real oil

production rate (Figure 8, upper). This caused the simulated total Dickman field

production rates to mismatch during the 1993-1998 production period of the Dickman A1

well (Figure 8, lower with dashed oval). A better history match is obtained by increasing

the Dickman A1 bottom hole depth about 9 feet (Figure 9, left), so that all four simulation

layers could be perforated for Dickman A1. This results in an improved result on the

total field oil production rate (Figure 9, right). This suggests that either the input

reservoir parameters, total volume, porosity, or permeability, need to be re-adjusted, since

the real production rate is from the reported perforation zone.

Several different capillary pressure models were tested in the simulation to study the

influence of the capillary pressure for transition zones. The simulation results indicate

that the capillary pressure almost had no influence either on the oil or water production

rate, so the capillary pressure effect was removed from the final calculation. Figure 10

shows the reservoir pressure distribution at the end of history matching simulation.

Work Plan for the Next Quarter No work plan as this is last quarter of the project.


11

Cost and Milestone Status Baseline Costs Compared to Actual Incurred Costs

2009 Jan 1 – Mar 31 Plan Costs Difference

Plan minus Costs Federal $25,000 $10,560 ($24,440)

Non-Federal $9,410 $0 ($9,410) Total $34,910 $10,440 ($33,850)

Forecasted cash needs vs. actual incurred costs

Notes: 1. Federal plan amount based on original award of $400K averaged over 12 reporting quarters. 2. Cost this period reflects 2 months salary for J. Zeng 3. Non-Federal plan amount based on original budget cost share of $150,573 averaged as above.

Actual Progress Compared to Milestones

Summary of Significant Events No problems to report.

Project Personnel

Prof. Christopher Liner is Principle Investigator and lead geophysicist. He is a member

of the SEG CO2 Committee, Associate Director of the Allied Geophysical Lab, and has

been selected to deliver the 2012 SEG Distinguished Instructor Short Course.


12

Dr. Jianjun (June) Zeng has been working exclusively on this project since Dec 2007 and

is lead geologist. She will be funded through the end of 2009.

Heather King is a graduate MS student in geophysics who joined the project in January

2009 as a research assistant. She will be funded out of the project Jan-May and Sept-

Dec 2009. Heather when she anticipates graduating. Her thesis will focus on Fort Scott

to demonstrate the integrity of this formation as a seal for injected CO2. This will

involve subtle structure and stratigraphy inferred by interpretation of multiple seismic

attributes.

Dr. Po Geng has been working on this project as a specialist consultant since February,

2009. He will be funded out of the project, considered part-time, through the end of 2009.

Technology Transfer Activities An invited lecture will be given at the Offshore Technology Conference in Houston, TX. Two posters will be presented at the DOE Annual CCS Conference in Pittsburgh, PA.

Contributors Christopher Liner (P.I, Geophysics)

Jianjun (June) Zeng (Geology and Petrel Modeling)

Po Geng (Flow Simulation)

Heather King (Geology and Geophysics, MS Candidate)


13

References

Blakely, R., 2004, Paleogeography and Geologic Evolution of North America, http://jan.ucc.nau.edu/%7Ercb7/nam.html , 2010.

Carr, T. R., D.F. Merriam, and J.D. Bartley, 2005, Use of relational databases to evaluate regional petroleum accumulation, groundwater flow, and CO2 sequestration in Kansas: AAPG Bulletin, 89, no 12, 1607–1627. Dandekar A. Y., 2006, Petroleum reservoir rock and fluid properties: CRC Press, 100. Hilpman P. L., M.O. Oros, D. L. Beene and E. D. Goebel, 1964, Oil and Gas Development in Kansas during 1963: State Geological Survey of Kansas Bulletin, 172. Li, C. F., and C. L. Liner, 2008, Wavelet-based detection of singularities in acoustic impedance from surface reflection data: Geophysics, 73, V1. Marfurt, K., 2006, Robust estimates of 3D reflector dip and azimuth: Geophysics, 71, 29. McCain, W.D. Jr., 1991, Reservoir-Fluid Property Correlations – State of the Art: SPERE, 6, no. 2, 266-272. Merriam, D. F., 1963, The Geologic History of Kansas, 1963: Kansas Geological Survey Bulletin, 162. Nissen, S. E., T. R. Carr, and K. J. Marfurt, 2006, Using New 3-D Seismic Attributes to Identify Subtle Fracture Trends in Mid-Continent Mississippian Carbonate Reservoirs: Dickman Field, Kansas: AAPG Search and Discovery, Article #40189. Ramaker, B. J., 2006, Influence of Mississippian Karst Topography on Deposition of the Cherokee Group: Ness County, Kansas, http://hdl.handle.net/1808/5544 , 2010.

Sawin, R. S., E. K. Franseen, R. R. West, G. A. Ludvigson, and W. L. Watney, 2008, Clarification and Changes in Permian Stratigraphic Nomenclature in Kansas: Kansas Geological Survey, Current Research in Earth Sciences Bulletin: 254, part 2, http://www.kgs.ku.edu/Current/2008/Sawin/index.html , 2010.

Zeller D. E. edit, 1968, The Stratigraphic Succession in Kansas: Kansas Geological Survey Bulletin, 189, http://www.kgs.ku.edu/Publications/ Bulletins/189/index.html , 2010.


14

Appendix A: Relative Permeability Equations Stone correlation formulations were used to generate the relative permeability model for

relative permeability of water, oil and water, gas and water, and gas. The relevant

equations are:

where

- Endpoint Saturation: Connate Water

- Endpoint Saturation: Critical Water

- Endpoint Saturation: Irreducible Oil for Water-Oil Table

- Endpoint Saturation: Residual Oil for Water-Oil Table

- Endpoint Saturation: Irreducible Oil for Gas-Liquid Table

- Endpoint Saturation: Residual Oil for Gas-Liquid Table

- Endpoint Saturation: Connate Gas

- Endpoint Saturation: Critical Gas

- at Connate Water

- at Irreducible Oil

- at Connate Liquid

- at Connate Gas

- Exponent for calculating from




15


The following parameters were used in the final calculation:

,

,

,

,

,

= 4


16

Tables

Table 1: Table of contents and assignments for planned final report of the Dickman study


17

Well name

Water Data Collection

Starting Time

Water Breakthrough

Time Production

Starting Date Production End Date

1 Dickman A1 N/A N/A 1/1/1993 2/28/2006

2 Humphrey 1 N/A N/A 1/1/1973 11/30/1994

3 Sargent 5 N/A N/A 9/1/1972 11/30/2005

4 Tilley 2 N/A N/A 11/1/1962 4/30/1998

5 Dickman 1 1/1/1962 1/1/1962 1/1/1962 1/31/2009

6 Humphrey 2 12/1/1994 12/1/1994 12/1/1994 1/31/2009

7 Humphery 3 10/1/2000 10/1/2000 10/1/2000 1/31/2009

8 Humphrey 4 4/1/2008 4/1/2008 4/1/2008 1/31/2009

9 Tilley 5 8/31/1997 8/31/1997 7/1/1995 1/31/2009

10 Dickman 2 1/1/1993 1/1/1993 9/1/1962 1/31/2009

11 Dickman 3 1/1/1993 1/1/1993 11/1/1962 1/31/2009

12 Dickman 6 1/1/1993 1/1/1993 11/1/1976 1/31/2009

13 Elmore 1 1/1/1993 1/1/1993 2/1/1963 6/30/2002

14 Phelps 1a 1/1/1993 1/1/1993 6/1/1977 1/31/2009

15 Tilley 1b 1/1/1993 1/1/1993 11/1/1962 1/31/2009

Table 2: Production starting date, ending date and water break through time for all 15 production wells in Dickman field. Notice only 5 wells (Dickman 1, Humphrey 2, Humphery 3, Humphrey 4 and Tilley 5) have a complete water production record.


18

Figures

Figure 1. Relationship between seismic amplitude and spice data. A) Base map showing vertical seismic line through selected wells. B) Amplitude section and formation tops. C) Spice data with tops. D) Spice data with amplitude wiggle trace overlay.

Figure 2. Top Mississippian time structure maps. (A) Tracking result using zero crossing in the amplitude volume. (B) Tracking result using peak in spice volume. Note improved continuity on this irregular, karst surface in the spice-generated map.


19

Figure 3. Relationship of spice to other attributes in vertical view along line in Fig 2. (A) Spice; (B) Instantaneous frequency; (C) Negative curvature; (D) Positive curvature.

Figure 4. Left: Water saturation from Dickman 2 well core measurement (other wells show similar range of water saturation). Right: Typical alterations in the fluid saturations of a core sample from virgin productive formation that was badly flushed with water-based drilling mud (from Dandekar, 2006). A 70% water saturation measured from flushed cores corresponds to 30% in-situ water saturation.


20

Figure 5. Chart to the left is the stratigraphic rank accepted by the Kansas Geological Survey (Sawin et al., 2008). The chart to the right shows correlation to Dickman local stratigraphic units, with higher confidence correlations in bold. The blue vertical bar indicates the target strata for Dickman research.

Figure 6: History matching flow simulation grid showing all 15 production wells and the Dickman 4 injection well.


21

Figure 7: History matching results for Dickman 1 production well (1962- 2009). A good match on oil and water production rate was obtained here.


22

Figure 8: Upper: Simulation result for Dickman A1 for all production years (1993-1998) using perforations in only the two top simulation layers. The simulated oil production rate (dash blue line) is much lower than the real rate (green dots). Lower: Dickman field total oil production rate shows significant mismatch in the 1993-1998 time period resulting from errors with the Dickman A1 well.


23

Figure 9. Upper: Improved match for oil production rate at Dickman A1 well after all four simulation layers are assumed to be perforated. Lower: Dickman field total daily oil production rate shows an improved match after all four simulation layers are perforated at the Dickman A1 well. To achieve this match, it was necessary to assume the Dickman A1 well had total depth six feet deeper than the reported value, or there was communication between the simulation layers.


24

Figure 10. Final reservoir pressure at the end of the history match simulation run.

Documents

Application of Cutting-Edge 3D Seismic Attribute ... · PDF fileApplication of Cutting-Edge 3D Seismic Attribute Technology to the Assessment of Geological Reservoirs for CO ... Geology